Figure 3

Distinct clustering of pleomorphic dermal sarcomas and potential protein markers to distinguish PDS from other skin malignancies. (a) t-SNE plot of malignant melanoma (MM), pleomorphic dermal sarcoma (PDS), cutaneous squamous cell skin cancer (cSCC) and skin normal tissue (SNT). The color represents the regional probabilities of being associated to one of the entities. (b) To identify proteins that differed specifically between cSCC and PDS, the data were normalized, scaling expression values between − 1 and 1. Then, a ranking process was employed based on two metrics: the chi-squared (χ²) statistic and the information gain ratio (IGR). The χ² statistic evaluated the observed frequencies of protein count differences between the groups, while the IGR quantified the information a protein provides about the differentiation between the groups of cSCC and PDS. Using these criteria, the top 50 genes showing the most pronounced differences in abundance were visualized using a heatmap. Columns corresponded to proteins and rows to individual samples. Human genome organization gene symbols are used for annotating the proteins (cSCC, n = 20; PDS, n = 24). (c) Boxplot comparing MM (n = 27), PDS (n = 24) and cSCC (n = 20) using a compared using a Wilcoxon Rank Sum Test. (d) Feature plot (UMAP) of single cell sequencing data¹² of skin normal tissue (SNT) for both MXRA8 and MAP1B. Clusters of both fibroblasts and pericytes are highlighted (originated from n = 5 samples of human skin tissue).